Sensory and chemical changes in Chilean Cabernet Sauvignon wines during storage in bottles at different temperatures

Transcription

1 Journal of the Science of Food and Agriculture J Sci Food Agric 81:1561±1572 online: 2001) DOI: /jsfa.971 Sensory and chemical changes in Chilean Cabernet Sauvignon wines during storage in bottles at different temperatures Hanne K Sivertsen,* Erik Figenschou, Frithjof Nicolaysen and Einar Risvik Norwegian Food Research Institute, Oslovn 1, N-1430 Ås, Norway Abstract: Two Chilean wines made from the Cabernet Sauvignon grape were stored in bottles at four different temperatures for a period of 44 months. During storage they were evaluated by sensory descriptive and chemical analyses. Different sensory pro les were obtained for the wines stored at the different temperatures. Smaller differences were found in the pro les of the two wines stored at the same temperatures. Wines were stored at different temperatures to determine if it was possible to predict the shelf-life of wines stored at optimal temperatures. This was not achieved in this study. Partial least squares regression was used to examine possible relationships between the chemical and sensory descriptive data. # 2001 Society of Chemical Industry Keywords: Chilean Cabernet Sauvignon; bottle storage; sensory; chemical; multivariate analyses INTRODUCTION Wine is subject to continuous changes in composition during storage. The rst change is generally towards higher quality, followed by a reverse development towards reduced quality. Throughout history, science has made great efforts to discover the `true' nature of wine, including the description of most compounds that impart the speci c characteristics of a wine aroma. For scienti c purposes the aroma is generally divided into three main groups: the aromas originating from the grape, the aromas produced during fermentation and the bouquet that results from the transformation of aromas during aging. 1±3 The aging bouquet can be subdivided into two types: the oxidative bouquet, which consists of acetaldehyde and acetals, and the reductive bouquet, which is developed under reductive conditions during aging in bottles. 4 Several studies have been performed in order to describe the chemical changes during bottle aging and their in uence on the sensory perception of the wines. Archier et al 5 studied the change in phenolic content of single-variety wines. Nagel and Wulf 6 found changes in the anthocyanins, avonoids and hydroxycinnamic acid esters of Merlot and Cabernet Sauvignon. Changes in volatiles, amino acids, organic acids, phenols and mineral content during storage have been studied extensively by Cantagrel and Carles. 7 Shinohara et al, 8 Shinohara and Watanabe, 9 Ough, 10 Rapp and GuÈnthert, 4 Ureta and Roblero, 11 Marais, 12 EtieÂvant et al 13 and Rapp and Marais 14 have all studied changes in volatiles during the storage of wine in bottles. Some of these studies also included sensory evaluation of the wines. 10±15 The purpose of this study was to evaluate bottled wines made from Cabernet Sauvignon grapes grown in Chile in terms of both sensory and chemical changes. These wines were stored for 44 months at four different temperatures. Wines were stored at accelerated temperatures to determine if it was possible to predict the shelf-life wines stored under optimal conditions. To examine possible interrelationships between sensory descriptions and chemical compounds, partial least squares regression PLS) 16 was performed. MATERIALS AND METHODS Wines The two Cabernet Sauvignon wines from Chile Canepa and St Digna) were stored in dark rooms at four different temperatures, 5, 10, 23 and 31 C. They were removed from the storage rooms 1day before the actual testing in order to bring them to room temperature. Descriptive analysis Eleven employees two females and nine males) at Vinmonopolet A/S Norwegian Wine and Spirits Monopoly), all regular participants in sensory quality panels, evaluated the wines in a sensory laboratory equipped with separate booths. All wines were evaluated under normal white light at room temperature 20 C). Six training sessions were held before the formal rating tests. Physical standards were used to assist in the de nition of the attributes. Fifteen attributes were * Correspondence to: Hanne K Sivertsen, 1709 Sycamore Ln, Davis, CA 95616, USA (Received 28 February 2000; revised version received 3 October 2000; accepted 20 August 2001) # 2001 Society of Chemical Industry. J Sci Food Agric 0022±5142/2001/$

2 HK Sivertsen et al selected on the basis of these training sessions Table 1). Eight aroma terms were chosen from the Wine Aroma Wheel. 17 Two visual attributes, colour intensity and colour hue, were also selected. Astringency and suppleness represented the mouthfeel characteristics. The remaining three attributes were bitterness, acidity and aging aroma. The two attributes suppleness and aging aroma were included in this study to describe the changes in the wines during storage. Suppleness can be described as softness on the palate and is developed in the wine during its shelf-life. Aging aroma is a complex, qualitative attribute. At the low end of the scale it is non-existent. At the middle of the scale it indicates the optimal quality of the wine. At the high end of the scale it is de ned as oxidised and maderized, with a high intensity of acetaldehyde. Each formal session lasted 1h. Sixteen samples 2 wines4 temperatures2 replicates) were served in a completely randomised order in clear tulip-shaped glasses marked with three-digit numbers and covered with Petri dishes. Tap water was used to rinse the mouth between samples, and all samples were expectorated. Owing to the long period of time between tastings, a training session with physical standards was held before each formal taste session. The samples were rated individually on an unstructured scale 150mm) ranging from `low intensity' to `high intensity', except for colour hue which ranged from `red/yellow' to `red/blue'. The results from the evaluations were converted into numbers by measuring the distance in centimetres from the left end of the scale to the marked intensity. Chemical analyses The wines were chemically analysed at the beginning of the study and after 18 months of storage. Ethanol, free and total sulphur dioxide, total acid, methanol, acetaldehyde, ethyl acetate, tartaric acid, succinic acid, malic acid, lactic acid, citric acid, acetic acid, glycerol, fructose, glucose, ph, colour intensity, colour hue and total phenol were determined by internal methods at A/S Vinmonopolet. Colour hue, colour intensity and total phenol 18 were analysed at the start of the storage period and also after 6 weeks, 3, 6, 12 and 18 months. Volatile compounds were analysed by coupled gas chromatography/mass spectrometry GC/MS). These analyses were performed at the beginning of the study and after each of the storage times mentioned above. An internal standard solution was prepared by dissolving methyl undecanoate g) in recti ed ethanol 100ml). Internal standard solution 50ml) was added to the samples 50ml), and the solution was extracted with dichloromethane Merck PA; 2 50ml), dried over sodium sulphate 10g) and ltered. The ltrate was reduced in a Rotavapor 1 to 10ml and further reduced to 1ml by blowing nitrogen over it. The samples were analysed using a Hewlett Packard 5890 Series II gas chromatograph coupled with a Hewlett Packard 5971A mass spectrometer, a Hewlett Packard 7673A controller unit and a Hewlett Packard 7673A autosampler. The integrator was a Hewlett Packard Chem Station Rev B0 DOS). The column was a Hewlett Packard HP-20M Carbowax 20M, 25m0.2mm, 0.2mm lm thickness). Split 1:20) injections were made at an injection temperature of 250 C and with an injection volume of 1.0ml. The carrier gas was helium at a linear ow rate of 25cms 1. The oven temperature was held at 40 C for 1min, then raised at 5 C min 1 to 230 C and held for 5min. The mass range was 25±350amu and the scan rate was 1scans 1. The compounds were quanti ed using the internal Table 1. Sensory attributes, definitions and physical standards Attribute Written de nition Physical standards Colour intensity Intensity of colour in the wine No physical standards Colour hue Hue ranging from red/yellow to red/brown No physical standards Fruit/berry aroma Blackcurrant, raspberry, cherry, plum, citrus fruit, jam, dried fruit Lemon:both juice and peel, a plum:canned red plum juice, blackcurrant jam, blackcurrant juice, liqueur of cherry before yeast was added Spicy aroma Black pepper, anise, mint Grained pepper, anise extract Vegetative aroma Eucalyptus, bell pepper, cooked vegetables cabbage, cauli ower, asparagus) Eucalyptus leaf left in glass of wine ca 15min, bell pepper green) left in glass of wine ca 15min, canned asparagus Woody aroma Vanilla, cedar tree Concentrate from the odour kit `le nez' Burned aroma Creosote, tar Concentrate from the odour kit `le nez' Floral aroma Rose, geranium Concentrate from the odour kit `le nez' Caramel aroma Burned sugar, butter No physical standards Chemical aroma Sulphur, acetic acid No physical standards Aging aroma Oxidised, maderised, acetaldehyde Oxidised wine Suppleness Mouthfeel of softness No physical standards Astringency Mouthfeel of dryness No physical standards Bitterness Bitter taste No physical standards Acidity Acidic taste No physical standards a The physical standards were added to neutral wine served in a wineglass J Sci Food Agric 81:1561±1572 online: 2001)

3 Changes in Chilean Cabernet Sauvignon wines during storage standard method. In the calculations, all response factors were assumed to be unity. Data analyses A two-way analysis of variance ANOVA) was run on the data from each examination time and each temperature, with the assessors as random effect. The ANOVA was performed by the General Linear Model GLM) procedure SAS Institute Inc, Cary, NC, USA). Tukey's test was used to determine signi cant differences among temperatures. Principal component analysis PCA) Unscrambler; Camo A/S, Trondheim, Norway) was performed on the sensory data. As validation procedure, cross-validation was used. The data were scaled to unit variance before calibration. Partial least squares PLS) Unscrambler) was used to examine possible interrelationships between sensory descriptions and chemical compounds. RESULTS AND DISCUSSION Effect of storage time on sensory attributes To evaluate a temperature effect on changes in sensory attributes, a two-way ANOVA was run after each of the eight sessions Tables 2 and 3). The attributes showing the largest changes during storage, namely colour hue, aging aroma and fruity aroma, are presented in Figs 1±3 respectively for Canepa and in Figs 4±6 respectively for St Digna. The attributes that did not show signi cant differences across temperatures were removed from the analyses. These were vegetative aroma and spicy aroma for Canepa and bitterness for both wines. For Canepa, no signi cant judge temperature effect was found during the rst three sessions, until 3 months of storage. From 6 until 33 months of storage, signi cant effects were found for most attributes and should be considered when evaluating the data Table 2). At 44 months of storage, signi cant effects were found for the attributes colour intensity, bitterness and acidity. For St Digna, a signi cant judge temperature effect was found at the start of storage for most attributes, and later for some attributes at each session, but no particular attributes. Further information can be obtained from the authors. Six weeks of storage For the most part there were no signi cant temperature effects. It is likely that the changes in chemical aroma for St Digna and in suppleness for Canepa were caused by factors other than temperature. Three months of storage For the Canepa wines stored at 23 C, the signi cant decrease in colour hue and increase in aging and chemical aromas may be due to bottle variation. The St Digna wines stored at 31 C showed signi cant changes in colour hue and aging aroma but not in other attributes. After 3 months, these wines stored at 31 C started deviating from the wines stored at lower temperatures. Table 2. Analysis of variance across temperatures for eight different storage times, 15 sensory attributes, 11 assessors, two replicates, Canepa Attribute Start 6 weeks 3 months 6 months 12 months 18 months 33 months 44 months Colour intensity NS NS NS * * NS * ** 10>31 5>23 31>23 31>23 Colour hue NS NS *** *** *** *** *** NS 5,31,10>23 5,10>23>31 5,10>23>31 5,10>23>31 5>10>23>31 Fruity NS NS NS *** *** *** *** ** 5,10,23>31 5,10>23,31 5,10>23,31 5>10>23,31 5>10,23,31 Woody NS NS NS * ** NS NS ** 31>5 31>5 5,31>10 Burned NS NS NS * ** NS ** NS 31>5 31>23,10,5 31>10,5 Floral NS NS NS * NS ** *** NS 5>31 5,10>31 5>23,31 Caramel NS NS NS NS ** *** ** NS 31>23,10 31>23,10,5 31>10,5 Chemical NS NS ** *** NS ** ** ** 23>10,5 31>10,5 31>10,5 31>23,10,5 10>23,31,5 Aging NS NS *** *** *** *** *** NS 23>31,10,5 31>23,10,5 31>23,10,5 31>23>10,5 31>23>10,5 Suppleness NS ** NS NS * ** NS *** 31>5 31>23 31>10,5 5,31>23,10 Astringency NS NS NS NS NS *** NS ** 5,10>23,31 23>31,5 Acidity NS NS NS ** NS NS NS ** 23,10,5>31 5>23,10 NS, not signi cant at p <0.05; *,**,*** signi cant at p <0.05, 0.01 and respectively. J Sci Food Agric 81:1561±1572 online: 2001) 1563

4 HK Sivertsen et al Table 3. Analysis of variance across temperatures for eight different storage times, 15 sensory attributes, 11 assessors, two replicates, St Digna Attribute Start 6 weeks 3 months 6 months 12 months 18 months 33 months 44 months Colour intensity NS NS NS NS * * NS *** 23>31 5>31 5,31>10,23 Colour hue *** NS *** *** *** *** ** NS 10>23,31,5 5>23,31 5>10,23>31 10,5>23>31 5>23,31/31>10,23 5>10>23>31 Fruity NS NS NS * ** ** *** ** 5>31 10,5>31 10,23,5>31 5,10>23,31 31,5>10 Spicy NS NS NS NS NS NS NS ** 31>5,23 Vegetative NS NS NS NS NS NS NS ** 31>5 Woody * NS NS NS * * ** NS 10>31,23,5 31>5 31>5 31>10,5 Burned NS NS NS NS *** NS ** ** 31>23,5,10 31>5 23>5 Floral NS NS NS NS * * ** ** Ð 10>31 10,5>31 31>5,23,10 Caramel NS NS NS NS ** NS ** ** 31>5 31>10,5 31,23,5>10 Chemical NS * NS NS * NS ** * 31>23 31>23,10 31>5 Aging NS NS * ** *** *** *** NS 31>10 31>23,5,10 31>23>10,5 31>23,10,5 31,23>10,5 Suppleness NS NS NS * NS *** ** ** 31>23,10 31>5,10,23 31>5 5>23,31 Astringency NS NS NS NS NS *** * ** 23,5,10>31 5>23 31,23>5 Acidity NS NS NS * NS * NS NS 10>31 23>31 NS, not signi cant at p <0.05; *,**,*** signi cant at p <0.05, 0.01 and respectively. Figure 1. Colour hue evaluated in Canepa stored at 5, 10, 23 and 31 C at start of storage period and after 6 weeks, 3, 6, 12, 18, 33 and 44 months of storage (11 judges2 replicates) J Sci Food Agric 81:1561±1572 online: 2001)

5 Changes in Chilean Cabernet Sauvignon wines during storage Figure 2. Aging aroma evaluated in Canepa stored at 5, 10, 23 and 31 C at start of storage period and after 6 weeks, 3, 6, 12, 18, 33 and 44 months of storage (11 judges2 replicates). Figure 3. Fruity aroma evaluated in Canepa stored at 5, 10, 23 and 31 C at start of storage period and after 6 weeks, 3, 6, 12, 18, 33 and 44 months of storage (11 judges2 replicates). J Sci Food Agric 81:1561±1572 online: 2001) 1565

6 HK Sivertsen et al Figure 4. Colour hue evaluated in St Digna stored at 5, 10, 23 and 31 C at start of storage period and after 6 weeks, 3, 6, 12, 18, 33 and 44 months of storage (11 judges2 replicates). Figure 5. Aging aroma evaluated in St Digna stored at 5, 10, 23 and 31 C at start of storage period and after 6 weeks, 3, 6, 12, 18, 33 and 44 months of storage (11 judges2 replicates) J Sci Food Agric 81:1561±1572 online: 2001)

7 Changes in Chilean Cabernet Sauvignon wines during storage Figure 6. Fruity aroma evaluated in St Digna stored at 5, 10, 23 and 31 C at start of storage period and after 6 weeks, 3, 6, 12, 18, 33 and 44 months of storage (11 judges2 replicates). Figure 7. Combined plot of scores and loadings for Canepa. PCA. PC1 (77%) and PC2 (7%). The labels represent the wines stored at four different temperatures (5=5 C, 10=10 C, 23=23 C and 31=31 C) for six storage intervals (1=start, 2=6 weeks, 3=3 months, 4=6 months, 5=12 months, 6=18 months, 7=33 months and 8=44 months). Figure 8. Combined plot of scores and loadings for St Digna. PCA. PC1 (58%) and PC2 (13%). The labels represent the wines stored at four different temperatures (5=5 C, 10=10 C, 23=23 C and 31=31 C) for six storage intervals (1=start, 2=6 weeks, 3=3 months, 4=6 months, 5=12 months, 6=18 months, 7=23 months and 8=44 months). J Sci Food Agric 81:1561±1572 online: 2001) 1567

8 HK Sivertsen et al Six months of storage After 6 months of storage, wines stored at 31 C showed a temperature effect for several sensory attributes. Both wines had decreased colour hue and increased aging aroma. Canepa wines stored at 31 C also showed increased woody, burned and chemical aromas. Twelve months of storage Almost all attributes were found to differ signi cantly for the two wines across the different temperatures. Major differences between Canepa and St Digna were found in the more pronounced loss of fruity aroma for Canepa stored at 23 and 31 C. The wines stored at 31 C had developed an oxidised aroma. Eighteen months of storage A signi cant increase in suppleness and decrease in astringency were found in the two wines stored at 31 C. The difference in colour hue for St Digna was not as great at 23 C as that for Canepa Figs 1 and 4). Increased aging aroma and decreased colour hue and fruity avours with increasing temperature have also been found in other studies. Rapp and Marais 14 found a signi cant increase in maturation bouquet for white wines stored at temperatures above 20 C. Another study on red wines showed a decrease in colour, aroma, taste and general quality with increasing temperature. 10 Thirty-three months of storage Aging aroma for all wines stored at 23 and 31 C had at this point reached a level on the scale where it was de ned as oxidised and maderised. The perceived intensity was stronger for Canepa than for St Digna Figs 2 and 5). The wines stored at 10 C showed a lesser increase, while those stored at 5 C were evaluated to have the same intensity as at 18 months. The wines stored at 23 C had not yet reached the same level as the wines stored at 31 C for the attributes aging aroma, colour hue and fruity/berry aroma. Forty-four months of storage After 44 months, all wines were evaluated to have more or less the same colour hue. Aging aroma showed different patterns between the two wines. At 5 C, this attribute for St Digna was perceived as signi cantly higher in intensity than at other temperatures, while for Canepa it was perceived as having the same intensity across all temperatures. Principal component analysis Principal component analysis was run on the average sets of data 11 assessors2 replicates) of both wines to examine attribute relationships and to show the shifts in wine sensory characters at different examination times and different temperatures. Figs 7 and 8 show the biplots based on the sensory evaluations of Canepa and St Digna respectively. For Canepa, principal component 1 PC1) and PC2 accounted for 77 and 7% of the variation in the data set respectively. For St Digna, PC1 and PC2 accounted for 58 and 13% respectively. For Canepa, the rst PC describes the development of aging characters in the wines during the storage period. PC2 depicts changes in colour intensity and mouthfeel. It can be observed that a majority of the wines are situated in the positive part of PC1. In this group the wines stored at 10 C for 33 months represent the wines stored the longest and which can still be considered to have maintained the same characteristics as the younger wines in the same group. Then, as the wines are located further along PC1, it can be concluded that these wines have developed different taste pro les, dominated by aging aroma. It can also be observed that the wine stored at 23 C for 3 months and the wine stored at 31 C for 6 weeks have maintained the same characteristics as wines stored at optimal temperatures until 33 months of storage for this particular wine. The wines are all located along PC1 except for the two stored at 10 and 23 C for 44 months. These wines seemed to have kept a higher intensity of astringency than the other wines. The change in pattern between the wines stored for 44 months and the other wines could be related to the two attributes aging aroma and colour hue. Until 33 months of storage the major differences among the wines were based on these two attributes. After 44 months of storage, no signi cant difference among the wines was found for these attributes, and the wines were evaluated based on other attributes. The mouthfeel attributes seem to be the most pronounced. For St Digna in general, a similar pattern can be seen. For Canepa, the rst principal component described the major variability of the wines, and PC2 could be interpreted as a mouthfeel dimension. For St Digna, this trend has been continued. The wines located along PC2 had developed a softer palate, still maintaining the characteristics from the beginning of storage. Only the wines stored at the higher temperatures, 23 and 31 C, had developed aging aromas connected to PC1. It can be seen that after 44 months of storage also the wines stored at 5 and 10 C had developed an aging character as found in the wines stored at higher temperatures. The St Digna wines that had kept the same sensory characteristics as the wines stored at optimal temperature, 10 C for 44 months, were the ones stored at 23 C for 18 months and at 31 C for 6 months. Possible relationships between sensory attributes and chemical compounds Partial least squares regression PLS2) was performed on the data from Canepa and St Digna to examine possible relationships between sensory attributes and chemical compounds. Only data from the rst J Sci Food Agric 81:1561±1572 online: 2001)

9 Changes in Chilean Cabernet Sauvignon wines during storage Figure 9. Combined plot of scores and loadings for Canepa. PLS. PC1 (53%) and PC2 (7%). The labels represent the wines stored at four different temperatures (5=5 C, 10=10 C, 23=23 C and 31=31 C) for six storage intervals (A=start, B=6 weeks, C=3 months, D=6 months, E=12 months and F=18 months). months of storage were included, as no chemical analyses were performed after this time. Figs 9 and 10 show the biplots of the results from Canepa and St Digna respectively. For Canepa, PC1 accounted for 53% and PC2 for 7% of the total explained variance of 87% in the validation model, after four interpretable principal components. For St Digna, PC1 and PC2 accounted for 63 and 6% respectively three interpretable PCs, total explained variation 81%). Tables 4 and 5 show the cumulative validated explained variance for each of the sensory attributes and chemical compounds from the data of Canepa and St Digna respectively. Eight sensory variables showed a low percentage of explained variance. These were colour intensity, woody aroma, burned aroma, oral aroma St Digna), chemical aroma St Digna), caramel aroma Canepa), astringency Canepa) and acidity. The chemical compounds lactic acid ethyl ester, 1- hexanol, acetic acid, 2,3-butanediol, diethyl succinate and 2-phenylethanol were identi ed by comparing the mass spectra and retention times with those from authentic compounds. 2-Hydroxy-3-butenone, 1,3- butanediol and butyrolactone were identi ed by Figure 10. Combined plot of scores and loadings for St Digna. PLS. PC1 (63%) and PC2 (6%). The labels represent the wines stored at four different temperatures (5=5 C, 10=10 C, 23=23 C and 31=31 C) for six storage intervals (A=start, B=6 weeks, C=3 months, D=6 months, E=12 months and F=18 months). J Sci Food Agric 81:1561±1572 online: 2001) 1569

10 HK Sivertsen et al Table 4. Cumulative validated explained variance for each sensory attribute and chemical compound after three principal components (PLS2), Canepa Variable Code PC1 PC2 PC3 Colour intensity C intens Colour hue C hue Fruity aroma Fruitb Spicy aroma Spicy Vegetative aroma Veget Woody aroma Tree Burned aroma Burned Floral aroma Floral Caramel aroma Caramel Chemical aroma Chemical Aging aroma Aging Suppleness Soft Astringency Astr Acidity Acidic Colour hue chemical) C hue Ch) Lactic acid ethyl ester Lacacest Butanediol Butanedi Diethyl succinate Dietsucc Unknown 1 Unknown Unknown 2 Unknown Unknown 3 Unknown comparing the mass spectra with those from the NBS Mass Spectra Database. The compounds that did not differ in intensity among the samples during the storage period, in addition to the samples where the standard deviation was greater than the variation, were removed. The remaining compounds are listed in Table 6. In Figs 9 and 10, most of the variability was explained by PC1. For Canepa, vegetative and spicy aromas were negatively correlated to lactic acid ethyl ester, `unknown 3' and, to a lesser extent, `unknown 2'. According to Table 6, lactic acid ethyl ester increased up to 12 months, then seemed to level out. A small increase caused by higher temperatures can be seen. This is in agreement with Shinohara et al, 8 who proposed an aging period for ester equilibrium of about 1year for lactic acid and more than 2 years for succinic acid. A similar pattern can be seen for St Digna, but included among the chemical compounds that were negatively correlated were also colour hue and 1,3- butanediol. This chemical compound and 2,3-butanediol were found to increase in intensity until 12 months, then decrease until 18 months of storage Table 6). No distinct differences among the samples stored at different temperatures can be seen. Cantagrel and Carles 7 found a decrease in these compounds in red wines after 6 years of storage, but the variation was not regular. Colour hue, fruity aroma, astringency and oral aroma for Canepa were negatively correlated to suppleness, aging aroma, chemical aroma, diethyl succinate and chemical colour hue. Marais 12 found in white wines during storage that diethyl succinate markedly increased at all temperatures 0, 10, 20 and 30 C). This is in agreement with what was found by other authors. 7,8 Rapp and GuÈnthert 4 suggested that this increase could be ascribed to chemical esteri cation, which begins at the end of fermentation. The negative correlation between sensory and instrumental colour hue is logical, since a browning process in the wine gives increased values instrumentally. High 420/520 nm values indicate a brown colour, while low values indicate a red colour. The clusters in PLS2 corresponded with those from PCA for both wines. One difference was noted for the wines stored for 6 months. In PLS they showed similar scores to those stored for 12 and 18 months and were grouped depending on the storage temperature. From PCA, only the wine stored for 6 months at 31 C 6±31) deviated from the wines stored for the same length of time, as seen by the score values of this wine. Thus the chemical analyses were more sensitive to changes, and the perceived changes seemed to appear later. CONCLUSIONS Descriptive sensory patterns in the two wines St Digna and Canepa were obtained during shelf-life. Storage at different temperatures gave different pro les, but it was not possible in this study to predict the shelf-life of wines stored at optimal temperatures. For future studies, GC-sniff would help in identifying and combining the compounds responsible for some of the most obvious perceptive changes in the wines. Table 5. Cumulative validated explained variance for each sensory attribute and chemical compound after three principal components (PLS2), St Digna Variable Code PC1 PC2 PC3 Colour intensity C intens Colour hue C hue Fruity aroma Fruitb Spicy aroma Spicy Vegetative aroma Veget Woody aroma Tree Burned aroma Burned Floral aroma Floral Caramel aroma Caramel Chemical aroma Chemical Aging aroma Aging Suppleness Soft Astringency Astr Acidity Acidic Colour hue chemical) C hue Ch) Lactic acid ethyl ester Lacacest ,3-Butanediol 1,3Butanedi ,3-Butanediol Butanedi Diethyl succinate Dietsucc Unknown 4 Unknown Unknown 1 Unknown Unknown 2 Unknown Unknown 3 Unknown J Sci Food Agric 81:1561±1572 online: 2001)

11 Changes in Chilean Cabernet Sauvignon wines during storage Table 6. Chemical compounds (mg l 1, except for colour hue) measured at different temperatures and storage times Storage time Canepa St Digna Compound 5 C10 C23 C31 CCompound 5 C10 C23 C31 C Start of storage Lactic acid 8.2 Lactic acid weeks ethyl ester ethyl ester months months months months Start of storage Butanediol 0.5 1,3-Butanediol weeks months months months months Start of storage Diethyl succinate 2.4 Butanediol weeks months months months months Start of storage Unknown Diethyl succinate weeks months months months months Start of storage Unknown Unknown weeks months months months months Start of storage Unknown Unknown weeks months months months months Start of storage Colour hue 0.6 Unknown weeks months months months months Start of storage Unknown weeks months months months months Start of storage Colour hue weeks months months months months ACKNOWLEDGEMENTS The authors wish to express their sincere gratitude to Mr Miguel Torres Soc Vinicola Miguel Torres SA, Chile) for the contribution of wine. We are grateful for all the help at A/S Vinmonopolet Norwegian Wine and Spirits Monopoly) from the tasters and lab personnel. Many thanks are due to Ulla Dyrnes Matforsk) for illustrations and to Mr Tore Hùyem J Sci Food Agric 81:1561±1572 online: 2001) 1571

12 HK Sivertsen et al Department of Food Science, Norwegian University of Agriculture,) for his support as a supervisor. REFERENCES 1 Ribereau-Gayon P, Wine avor, in Chemistry and Technology, Ed by Charalambous G and Inglett GE, Academic Press, New York, pp 355± ). 2 Rapp A and Mandery H, Wine aroma. Experimenta 42:873± ). 3 Rapp A, Wine aroma substances from gas chromatographic analysis, in Wine Analysis, Ed by Linskens HF and Jackson JF, Springer, Berlin, pp 29± ). 4 Rapp A and GuÈnthert M, Changes in aroma substances during the storage of white wines in bottles, in The Shelf Life of Foods and Beverages, Ed by Charalambous G, Elsevier Science, Amsterdam, pp 141± ). 5 Archier P, Coen S and Roggero J-P, Changes in phenolic content of single-variety wines after the rst nine months of storage. Sci Alim 13:483± ). 6 Nagel CW and Wulf W, Changes in the anthocyanins, avonoids and hydroxycinnamic acid esters during fermentation and aging of Merlot and Cabernet Sauvignon. Am J Enol Vitic 30:111± ). 7 Cantagrel R and Carles J, CaracteÂrisation analytique des vins de ceâpage et pheânomeánes de vieillissement. Ann Nutr Alim 32:1073± ). 8 Shinohara T, Shimizu J and Shimazu Y, Esteri cation rates of main organic acids in wines. Agric Biol Chem 43:2351± ). 9 Shinohara T and Watanabe M, Effects of fermentation conditions and aging temperature on volatile ester contents in wine. Agric Biol Chem 45:2645± ). 10 Ough CS, Some effects of temperature and SO 2 on wine during simulated transport or storage. Am J Enol Vitic 36:18± ). 11 Ureta FC and Roblero VC, Evaluacion quimica y organoleptica de vinos tintos nos Cabernet Sauvignon), de differentes edades, pero de un mismo origen. Agric Tecn Chile) 45:241± ). 12 Marais J, Effect of storage time and temperature on the volatile composition and quality of South African Vitis vinifera L cv Colombar wines, in The Shelf Life of Foods and Beverages, Ed by Charalambous G, Elsevier Science, Amsterdam, pp 169± ). 13 EtieÂvant PX, Issanchou S, Marie S, Ducruet V and Flanzy C, Sensory impact of volitile phenols on red wine aroma: in- uence of carbonic maceration time of storage. Sci Alim 9:19± ). 14 Rapp A and Marais J, The shelf life of wine: changes in aroma substances during storage and aging of white wines, in Shelf Life Studies of Foods and Beverages. Chemical, Biological, Physical and Nutritional Aspects, Ed by Charalambous G, Elsevier Science, Amsterdam, pp 891± ). 15 Stern DJ, Guadagny D and Stevens KL, Aging of wine: qualitative changes in the volatiles of Zinfandel wine during two years. Am J Enol Vitic 26:208± ). 16 Martens H and Nñs T, Multivariate Calibration. Wiley, Chichester 1989). 17 Noble AC, Arnold RA, Buechsenstein J, Leach EJ, Schmidt JO and Stern PM, Modi cation of a standardized system of wine aroma technology. Am J Enol Vitic 38:143± ). 18 Singleton VL and Rossi Jr JA, Colorimetry of total phenolics with phosphomolybdic±phosphotungstic acid reagents. Am J Enol Vitic 16:144± ) J Sci Food Agric 81:1561±1572 online: 2001)